On the blind separability of multiple user signals in the presence of delay spread

We study the problem of separating blindly multiple user signals at the output of a linear multiple-input-multiple-output (MIMO) system. A communication application is the simultaneous transmission of synchronous signals through a linear channel that introduces both intersymbol (ISI) and co-channel (CCI) interference. The signals are assumed to be non-Gaussian, i.i.d., and share the same statistical properties. We establish a set of necessary and sufficient conditions for perfect blind recovery of all the signals. Based on these conditions we propose and analyze some optimization criteria that lead to efficient blind adaptive multiuser separation/equalization algorithms.